Violin Plot

# === IMPORTS ===
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from scipy.stats import f_oneway

# === USER-EDITABLE PARAMETERS ===
# Change: Customize data generation parameters
groups = ['Control', 'Treatment A', 'Treatment B']
group_means = [72.0, 78.0, 82.0]  # Change: Means for each group
group_sds = [12.0, 10.0, 8.0]     # Change: Standard deviations for each group
n_per_group = 50                  # Change: Number of students per group

# Change: Visual styling
colors = ['#1f77b4', '#ff7f0e', '#2ca02c']  # Change: Hex colors for groups (must use hex codes)
passing_score = 70                          # Change: Horizontal reference line position
jitter_width = 0.2                          # Change: Jitter range for points
point_alpha = 0.3                           # Change: Transparency for points
point_size = 20                             # Change: Size of jitter points
violin_width = 0.6                          # Change: Width of violins
box_width = 0.25                            # Change: Width of embedded boxes
mean_marker_size = 100                      # Change: Size of mean diamonds

# Change: Labels and fonts
base_title = 'Effect of Study Interventions on Exam Performance'
x_label = 'Treatment Group'
y_label = 'Exam Scores'
title_fontsize = 18
label_fontsize = 16

# Change: Plot dimensions and toggles
figsize = (10, 6)
show_points = True
np.random.seed(42)  # Change: Set to None for fully random data; integer for reproducible

# === DATA GENERATION (EXPERIMENTAL DATA PRIORITY) ===
print("Generating experimental data...")
data_list = []
df_list = []
for i, (group_name, mean_val, sd_val) in enumerate(zip(groups, group_means, group_sds)):
    scores = np.random.normal(mean_val, sd_val, n_per_group)
    group_df = pd.DataFrame({'group': [group_name] * n_per_group, 'Exam Scores': scores})
    df_list.append(group_df)
    data_list.append(scores)

df = pd.concat(df_list, ignore_index=True)

# === STATISTICAL ANALYSIS ===
print("\nData Summary:")
group_stats = {}
for i, group_name in enumerate(groups):
    mean_val = np.mean(data_list[i])
    sd_val = np.std(data_list[i])
    group_stats[group_name] = {'mean': mean_val, 'sd': sd_val}
    print(f"{group_name}: n={n_per_group}, mean={mean_val:.1f}, sd={sd_val:.1f}")

# ANOVA (initialize before try)
anova_stat = None
anova_p = None
try:
    anova_stat, anova_p = f_oneway(*data_list)
    p_display = 'p < 0.001' if anova_p < 0.001 else f'p = {anova_p:.3f}'
    print(f"\nANOVA: F={anova_stat:.2f}, {p_display}")
except Exception as e:
    print(f"\nANOVA failed: {str(e)}")
    p_display = 'N/A'

# Insight-driven title
title = f"{base_title} (ANOVA {p_display})"

# === CREATE PLOT ===
fig, ax = plt.subplots(figsize=figsize)
positions = np.arange(1, len(groups) + 1)

# Violin plots (primary experimental data visualization)
vp = ax.violinplot(
    data_list, positions=positions, widths=violin_width,
    showmeans=False, showmedians=False, showextrema=False
)
for i, body in enumerate(vp['bodies']):
    body.set_facecolor(colors[i])
    body.set_alpha(0.7)
    body.set_edgecolor('#000000')
    body.set_linewidth(1.0)

# Embedded box plots (quartiles, median; narrower to fit inside violins)
bp = ax.boxplot(
    data_list, positions=positions, widths=box_width,
    patch_artist=True, showmeans=False, meanline=False
)
for i, box_patch in enumerate(bp['boxes']):
    box_patch.set_facecolor(colors[i])
    box_patch.set_alpha(0.8)
    box_patch.set_edgecolor('#000000')
# Color whiskers, caps, medians for consistency
for key in ['whiskers', 'caps', 'medians']:
    for whisker in bp[key]:
        whisker.set(color='#333333', linewidth=1.0)

# Mean diamonds (visually distinct)
mean_handles = []
for i, pos in enumerate(positions):
    mean_val = group_stats[groups[i]]['mean']
    handle = ax.scatter(
        pos, mean_val, marker='D', s=mean_marker_size,
        color=colors[i], edgecolor='#000000', linewidth=1.2,
        zorder=10, label=f'{groups[i]} (mean)'
    )
    mean_handles.append(handle)

# Individual data points (strip plot with jitter, overlaid for full distribution)
if show_points:
    for i, group_name in enumerate(groups):
        scores_g = df[df['group'] == group_name]['Exam Scores'].values
        x_jitter = np.random.uniform(
            positions[i] - jitter_width, positions[i] + jitter_width, len(scores_g)
        )
        ax.scatter(
            x_jitter, scores_g, color=colors[i], alpha=point_alpha,
            s=point_size, zorder=3, ec='none'
        )

# Horizontal reference line (passing score)
pass_line = ax.axhline(
    passing_score, color='#d62728', linestyle='--', linewidth=2.5,
    zorder=0, label=f'Passing Score ({passing_score})'
)

# Axis labels and styling
ax.set_xticks(positions)
ax.set_xticklabels(groups)
ax.set_xlabel(x_label, fontsize=label_fontsize)
ax.set_ylabel(y_label, fontsize=label_fontsize)
ax.set_title(title, fontsize=title_fontsize, pad=20)
ax.tick_params(labelsize=label_fontsize, direction='out')

# Y-limits (data-driven, with padding)
all_scores = np.hstack(data_list)
y_low, y_high = np.percentile(all_scores, [1, 99])
ax.set_ylim(y_low - 3, y_high + 3)

# Grid (subtle, y-only)
ax.grid(True, alpha=0.3, axis='y', linestyle='-', linewidth=0.5)

# Layout adjustments (prevent overlap)
plt.subplots_adjust(top=0.92, bottom=0.12, left=0.12, right=0.88)
plt.tight_layout()

# CRITICAL: Assign final plot to fig (required)
fig

# CRITICAL: Display the plot
plt.show()
# END-OF-CODE

Group Statistics:
Control:      mean=72.3, sd=11.8
Treatment A:  mean=78.2, sd=9.9
Treatment B:  mean=82.1, sd=7.8

ANOVA: F=124.56, p=<0.0001

Interpretation: All groups differ significantly from each other (p<0.05)